Your smartphone is already an extension of your body. It knows you — it understands you, right? But what if it knew how you were really feeling? Microsoft Research has built a prototype system that aims to detect the user’s mood. If the research pans out, phones in the future could add mood detection to the array of accelerometers, gyroscopes, and GPS chips that already customize the experience — emoticons could go the way of the dinosaurs.
Rather than go with a mood ring analog and develop some borderline pseudo-scientific physical sensor, Microsoft Research created a software package called MoodScope. This is a lightweight application that runs on a smartphone and uses context clues to infer how the user is feeling.
MoodScope measures a huge array of interactions with the phone including app usage, phone calls, emails, text messages, browsing history, and geographic location. This works out to thousands of data points each day. The team recruited 32 study participants and had them use the system for two months. During that time, they also completed self-assessments of their mood to see how close the software was getting to the truth.
The results of the preliminary study were shockingly (and maybe disconcertingly) accurate. Without any personal tuning, the app was able to guess at someone’s mood with 66% accuracy. That means anyone picking up the phone would see that level of exactness. When the software was “calibrated” for an individual user, the accuracy jumped as high as 93%.
MoodScope uses a two-axis scale to measure mood. On the X-axis is pleasure, and activeness is on the Y. This is an oversimplification of emotional states, but it can encompass general feelings like happiness, calm, and boredom. That’s probably all the detail you need to enhance a mobile experience, Microsoft says.
It is believed this technology could be used to improve interaction and recommendation online. A social network could automatically include your predicted mood, or a recommendation engine like Netflix could more accurately tailor itself to your state of mind at the time. If MoodScope proves to be a success, you might never again have to wonder if that text message you received was supposed to be sarcastic or serious.
The prototype build of MoodScope was developed for iOS and Android devices. Despite gathering and analyzing all that data, it only used a smidge of battery in the study — just 3.4 milliwatt-hours per day. That works out to 20 minutes of standby time on the average smartphone. It might be even less draining in practice as some devices running Android gather similar data for Google Now search. If MoodScope could plug into the vast stores of data Google already has, it might be even more effective. There are obvious privacy implications here, but we’ve adapted to worse.
MoodScope won’t be able to detect sudden swings in mood, or external factors that might cause stress. However, it may make your smartphone feel a little more personal. Work is continuing on the app in hopes it will be a real product someday.